Funnel optimization: Find and fix the stage costing you customers

Funnel optimization is the practice of measuring how many people move from one stage of a funnel to the next, identifying the stage with the largest drop-off relative to what's achievable, and making changes at that stage to fix the issues and lift conversions.

Today, most funnels aren’t built around a single webpage. That funnel includes what happens between visits as well as what happens on a page. A customer might browse a jacket on their phone, receive a reminder email, and ultimately buy it on their laptop days later. They might even try it on in-store or ask a friend for advice, actions without a digital footprint. By connecting the touchpoints you can track, you can unify these fragmented actions into a single customer journey. This clarity allows you to pinpoint exactly which efforts, whether an email, push notification, or SMS message, drove the conversion.

If you’re a team with a funnel already up and running but aren’t seeing the results you expected, this article will help you optimize your marketing funnel and measure the results.

TL;DR

• Funnel optimization measures how many people move between stages, finds the stage losing the most relative to what’s achievable, and changes what happens there. A funnel is a model, so one drawn only around a website will hide losses that happen off it.
• Each stage of the funnel, from awareness through consideration, conversion, and retention, loses people for different reasons. Diagnose the cause at each stage before choosing a fix, because the wrong fix can make things worse.
• The conversion funnel covers a single visit, while the engagement funnel follows the same customer from one visit to the next. A funnel scoped to sessions counts a returning customer as a new entrant, so retention looks lower and losses at the top look bigger.
• To know whether a change worked, measure a fixed cohort, choose the attribution window deliberately, compare against a holdout group, and state the denominator on every rate.
• Braze connects a unified customer profile, native execution across channels, and control groups, so teams can see each customer’s full journey, change a stage without an engineering ticket, and measure the lift.

What is funnel optimization?

Funnel optimization is the practice of measuring how many people move from one stage of a funnel to the next, identifying the stage with the largest drop-off relative to what's achievable, and making changes at that stage to fix the issues and lift conversions.

A funnel is a model of how customers move toward a goal, typically broken into these stages: awareness, consideration, conversion, and retention. Each stage is defined by what the customer has done, like viewing a product, starting a trial, or making a second purchase, regardless of the channel they used. Customers don’t always move in a straight line, so the business decides where each stage starts and ends and which actions and channels count. A funnel drawn only around website activity will hide valuable information.

Funnel optimization is about finding the stage of the full funnel where customers are dropping off, but it often gets confused with two related (and complementary) practices:

• Conversion rate optimization improves the share of people who complete a specific action on a given page or flow, such as a sign-up form or checkout.
• Funnel automation builds the workflows that move customers from stage to stage automatically, triggered by what they do.

How to optimize a marketing funnel in six steps

Here’s a quick overview of what the optimization process looks like:

1. Define the stages, including the ones that happen away from your website or app.
2. Measure how many people move from each stage to the next.
3. Find the stage with the largest loss relative to what’s achievable.
4. Diagnose the cause of the drop-off at that stage.
5. Change what happens at that stage.
6. Measure the result, and repeat.

Marketing funnel optimization: The four stages, and what moves each one

Marketing funnel optimization works one stage at a time. A typical awareness-to-conversion funnel loses people at several points, often for different reasons, so treating every drop-off with the same fix is a mistake. For example, people might leave after awareness because ads are reaching the wrong audience, or at checkout because of unexpected shipping costs. Knowing how to optimize a marketing funnel starts with finding the cause at each stage, then choosing funnel optimization strategies that address it.

Awareness

Awareness is where someone first learns about a brand and enters the funnel, whether that’s through a site visit, an app install, or a sign-up.

What to measure: How many people take a first action, or micro-conversion, such as signing up, downloading content, or opting in to email, push, or SMS.

Why people drop off here: What people see before they arrive doesn’t match what they find, or the permission request comes before the person has a reason to say yes.

Type of fix: Aligning the message with the destination, and timing permission requests so they make sense to the person being asked.

Consideration

Consideration is where someone who recognizes the brand starts weighing it by browsing, comparing, saving items, or starting a trial.

What to measure: Micro-conversions that show growing intent, such as product views, saves, wishlist adds, or trial starts, and how many people come back for a second visit.

Why people drop off here: The first follow-up message arrives at the wrong moment or says the wrong thing, or the person needs information the brand hasn’t given them.

Type of fix: Addressing the timing and relevance of follow-up messages, and providing content that answers the questions holding people back. Consider that if the first message is mistimed, adding additional messages may raise unsubscribe rates.

Conversion

Conversion is the step where intent becomes action, where someone makes a purchase, subscribes, or books a service.

What to measure: Stage conversion from the last point of intent, like an add to cart or a checkout start, to the completed action. For checkout, that means tracking checkout conversion rate.

Why people drop off here: Friction in the flow, such as unexpected extra costs, a long or complicated checkout, being asked to create an account, or a lack of trust at the point of decision.

Type of fix: Removing the friction point to rebuild trust in the checkout flow. Cart recovery via a well-timed abandoned cart email or push notification is one of the most widely used fixes for cart abandonment at this stage.

Retention

Retention covers everything after the first conversion. For apps and subscriptions, it also includes activation, the first point at which a new customer begins to see real value from a product.

What to measure: How many customers in each cohort are still active at set points after their first conversion. For apps and subscriptions, also track activation rate and time to first value.

Why people drop off here: New customers take too long to reach first value, or customers who drift away are never re-engaged.

Type of fix: Onboarding that shortens time to first value, and messaging matched to each customer’s lifecycle stage, including re-engagement for people who’ve gone quiet. Marketing funnel automation lets those messages trigger from what customers actually do.

Optimizing the conversion funnel versus optimizing the journey between visits

Two funnels run alongside each other, often measured by different teams and tools.

Conversion funnel optimization: improving the single visit

The conversion funnel follows a customer through a single session, from arriving on a site or app to completing an action like a purchase or sign-up. Conversion funnel optimization improves that path so more people finish what they started.

The engagement funnel: the journey between visits

An engagement funnel tracks how people interact with a brand across the whole relationship, from awareness through conversion and into retention and loyalty. It follows customers from one visit to the next, including the time in between. Most reported funnel losses are people who left with the intention to return, but were never brought back.

For example, say a shopper visits your site on Monday, adds a pair of boots to their cart, and leaves. On Thursday, a re-engagement email brings them back, and they buy the boots. A funnel scoped to sessions counts these as two separate entrants: one who dropped off at the cart, and one new entrant who converted. This makes retention look lower than it is, and the loss at the top of the funnel looks larger.

Session-based analytics and conversion-rate tools provide an accurate picture of what happens within a visit. Following customer actions and the session brings the time between visits into view, so it can be measured and improved through customer journey optimization.

How to measure funnel performance

The clearest way to understand funnel performance is to measure it the same way before and after you make a change. There are three ways to do this: by measuring a fixed cohort, deliberately defining your attribution window, and using a holdout group for comparison.

Measure a fixed cohort

A cohort is a fixed group of people who entered the funnel during the same period, such as everyone who signed up in the first week of a month. Following that one group through each stage shows how they progress. However, a funnel measured on whoever happens to be in it at any given moment mixes new and previous entrants.

Choose the attribution window deliberately

The attribution window is the amount of time after a message or change still counts toward influencing a conversion. A short window flatters within-session tactics, because their effect shows up right away, but it penalizes between-visit ones, like a reminder that brings someone back several days later, because the conversion lands after the window has closed.

Use a holdout group to attribute the change

A holdout group is a randomly selected set of people who don’t receive a change. Comparing their results with everyone else’s shows how much movement a change influenced, and how much came from seasonality, a promotion, or anything else happening at the time.

State the denominator on every rate

Stage conversion and other funnel rates are quoted against sessions, against the people who entered the funnel, and against the full addressable audience, and the three can produce very different numbers for the same funnel. Take 60 purchases from 1,000 sessions, made by 600 people, out of an audience of 10,000. That’s a 6% conversion rate per session, 10% per person who entered the funnel, and 0.6% of the addressable audience. Whenever you report on funnel performance, make it clear which denominator each rate is measured against and keep it consistent. The same applies to channel-level email metrics like click and conversion rates.

How Braze supports funnel optimization strategies

Braze measures and acts on the funnel from a unified customer profile, so the stage where people drop off and the message that might bring them back are managed in the same place. Braze Data Platform resolves each customer’s activity across app, web, and email into one profile, so a funnel can be measured per person instead of per session. A shopper who leaves on Monday and returns on Thursday remains one customer on one journey, and the return shows up as a return.

That profile also powers native execution across channels. Email, push, in-app messages, web, SMS, and WhatsApp all run from the same platform, so when a stage is losing people, the fix can reach each customer on the channel they’re most likely to respond to, without passing data between separate tools.

Because the data and channels sit together, ease of use extends to the funnel itself. Journey orchestration uses a no-code, drag-and-drop builder, so marketers can change what happens at a stage without an engineering ticket.

Once a stage has changed, lift and control groups show whether the change caused the improvement. Teams can run control groups for individual campaigns, or set up a Global Control Group that holds back a random share of customers from all messaging.

Grubhub used this approach when students started the onboarding process for its Campus program but weren’t finishing. To help them complete the process, the team built a 30-day welcome journey across email and push and tested different offers along the way. Grubhub+ Student activations among students in the journey were 188% higher than in the control group, and the campaign’s return on investment rose by 836%.

Within those journeys, BrazeAI™ handles timing and content selection at each lifecycle stage. Intelligent Timing sends each message when that person is most likely to engage, and item recommendations use past behavior to choose the products or content each customer sees.

As the enterprise-scale option, BrazeAI Decisioning Studio™ goes a step further, making 1:1 decisions on the content, offer, channel, and timing for each customer to optimize any business KPI.

See how Braze helps brands find and fix the stage that’s costing them customers.